Victoria Police Search Data

This dataset, which contains the summary of over 50,000 searches carried out by Victoria Police in the years 2018, 2019, 2022 and 2023, was prepared as part of The Racial Profiling Data Monitoring Project, a project of the Centre Against Racial Profiling.

This dataset took significant financial resources to obtain and transform into a user-friendly format. To find out more about how this dataset was obtained go to Annual Data. You can donate to the project here.

Note that 2020 and 2021 data is not included in this data due to the extraordinary circumstances of the COVID 19 pandemic, which are not a sound basis for inference regarding behaviour during non-pandemic periods.

Warrantless searches in scope

There are five types of searches without a warrant that are in scope for this dataset, summarised below. These represent searches that can be undertaken where an officer has ‘reasonable grounds’ to suspect that a person was in possession of a prohibited item specified as part of that search power.

  • Drugs: under the section 82 of the Drugs Poisons and Controlled Substances Act 1981
  • Volatile substances for inhalation for use by a person under 18: under sections 60E and 60F of the Drugs Poisons and Controlled Substances Act 1981
  • Weapons or dangerous articles: under Section 10 Control of Weapons Act 1990
  • Firearms: under section 149 of the Firearms Act 1996
  • Graffiti implements at a designated place: under section 13 of the Graffiti Prevention Act

Almost 90 percent of searches were for drugs. Note that some individuals were searched under multiple powers (eg where the officer stated that they had reasonable grounds to suspect the person had both weapons and drugs).

Ethnic appearance categories

Victoria Police records the ethnic appearance of the people that they search without warrant based on existing categories. These categories are not based on actual ancestry, but on perceived ethnicity as considered by the officer. The collection of this data has been mandatory since 2019.

Missing data

A total of 22,117 search records, or 42% of all records, did not have any ethnic appearance data entered, despite this field being mandatory. Records where the officer was unable to determine the racial appearance were marked as ‘Other’.

Middle Eastern appearance data limitations

There is a limitation on our ability to analyse police-targeting of people of Middle Eastern appearance, as in 2020 Victoria Police aggregated Mediterranean and Middle Eastern into a single category. This presents a challenge to our analysis, as we would ordinarily expect persons of Mediterranean descent (eg people of Italian, Greek, Croatian ancestry) to be categorised as ‘non-racialised’ or ‘White’. Meanwhile we would ordinarily expect persons of the Middle Eastern appearance to be considered ‘racialised’. The Centre Against Racial Profiling urges Victoria Police to separate ‘Middle Eastern’ from ‘Mediterranean’ in order for racial appearance data to be more useful. This is particularly relevant for understanding the effect of the impact of overseas conflicts and resulting protests at home on potential police bias.

Aboriginal and Torres Strait Islander appearance data caution

Please be reminded that perceived ethnicity data does not enable analysis of search data for police perception white appearing Aboriginal people. This and other ethnic groupings are subjective.

Summary statistics

The variables of interest in the dataset are summarised in Table 1 below.

Table 1 Characteristics of searches

Variable Stats / Values Freqs (% of Valid) Graph Missing
Year [factor]
1. 2022
2. 2019
3. 2018
4. 2023
14987(28.5%)
13281(25.2%)
12841(24.4%)
11545(21.9%)
0 (0.0%)
Racial.appearance [factor]
1. White
2. Middle Eastern/Med
3. Asian
4. African
5. Aboriginal
6. Other
7. Pacific Islander
8. South Asian
9. South American
19325(63.3%)
3999(13.1%)
2190(7.2%)
1677(5.5%)
966(3.2%)
785(2.6%)
776(2.5%)
704(2.3%)
115(0.4%)
22117 (42.0%)
Gender [factor]
1. M
2. F
3. U
30106(81.2%)
6965(18.8%)
17(0.0%)
15566 (29.6%)
Area.type [factor]
1. Metro
2. Regional
28423(73.2%)
10401(26.8%)
13830 (26.3%)
Unit.type [factor]
1. Uniform
2. Public Order Response
3. Transit
4. CIU
5. DRU
6. Other
7. PSO
8. Highway Patrol
37893(72.0%)
4673(8.9%)
3755(7.1%)
1979(3.8%)
1737(3.3%)
1093(2.1%)
958(1.8%)
566(1.1%)
0 (0.0%)
Contact.Type [factor]
1. P
2. V
31870(65.7%)
16622(34.3%)
4162 (7.9%)
Rank.of.Member [factor]
1. CONST
2. SCONST
3. PSO
4. SGT
5. SSGT
6. RECRUT
7. SUPT
29114(55.3%)
17968(34.1%)
3041(5.8%)
2431(4.6%)
93(0.2%)
6(0.0%)
1(0.0%)
0 (0.0%)
Found [factor]
1. No
2. Yes
43602(82.8%)
9052(17.2%)
0 (0.0%)

Generated by summarytools 1.0.1 (R version 4.4.1)
2024-09-25

The breakdown of searches by search type appears in Table 2 below. This has been summarised separately from the main data as some searches involved multiple search types - for example where an officer believed that they had reasonable grounds for suspecting a person was in possession with both drugs and weapons.

Table 2 Legislative basis for searches

Variable Stats / Values Freqs (% of Valid) Graph Missing
Search.type [factor]
1. Drugs
2. Weapons
3. Graffiti
4. Firearms
5. Volatile substance
47067(87.9%)
4611(8.6%)
982(1.8%)
815(1.5%)
41(0.1%)
0 (0.0%)

Generated by summarytools 1.0.1 (R version 4.4.1)
2024-09-25

Hit rate analysis

The ‘hit rate’ is the percentage of searches which resulted in the officer finding contraband as part of the search. The average hit rate across all searches was 17.2%.

Hit rates by ethnic appearance

Hit rates are lower for people who appeared to be African (12.9%), Middle Eastern/Mediterranean (13.8%), Asian (13.7%), South Asian (13.9%) and Aboriginal and Torres Strait Islander (15.6%) than for people who appeared White (18%).

A lower hit rate means that a lower percentage of searches of people in the group are resulting in the discovery of contraband. That is, the suspicion directed at groups with lower hit rates is more likely to be informed by racial bias, either conscious or unconscious. This may be evidence of racial targeting.

Table 3 Hit rates by variables of interest

label

variable

Found

Total

test

No

Yes

Racial.appearance

Aboriginal

815 (84.4%)

151 (15.6%)

966 (1.8%)

p value: <0.0001
(Pearson's Chi-squared test)

African

1460 (87.1%)

217 (12.9%)

1677 (3.2%)

Asian

1889 (86.3%)

301 (13.7%)

2190 (4.2%)

Middle Eastern/Med

3448 (86.2%)

551 (13.8%)

3999 (7.6%)

Missing

18145 (82.0%)

3972 (18.0%)

22117 (42.0%)

Other

661 (84.2%)

124 (15.8%)

785 (1.5%)

Pacific Islander

630 (81.2%)

146 (18.8%)

776 (1.5%)

South American

97 (84.3%)

18 (15.7%)

115 (0.2%)

South Asian

606 (86.1%)

98 (13.9%)

704 (1.3%)

White

15851 (82.0%)

3474 (18.0%)

19325 (36.7%)

Total

43602 (82.8%)

9052 (17.2%)

52654 (100.0%)

Figure 1 below shows how the ht rate for different racial appearances changed over the four year period, with a reference line showing the average hit rate for all searches over four years in red. The following observations are made:

  • People who appear White experienced higher hit-rates compared with the overall population in most years, except for 2022. This may be due to the effect of pandemic-related measures.
  • People of African, Middle Eastern/Mediterranean and Asian appearance experienced persistently lower hit rates than the average.
  • Hit-rates for people appearing to be Aboriginal and Torres Strait Islander have reduced over the four year period.

Figure 1 Changes in hit rates by racial appearance over time and compared to the average

Hit rates by other search characteristics

Other characteristics which may have influenced hit rates are summarised in Table 4 below. Different search types had widely different hit rates, with drug-related searches having the lowest hit-rate (15.2%) and graffiti-related searches at designated places, such as train stations, having the highest hit-rate (50.1%), followed by weapons-related searches (32.8%).

Table 4 Hit rates by other search characteristics

label

variable

Found

Total

test

No

Yes

Search.type

Drugs

39936 (84.8%)

7131 (15.2%)

47067 (87.9%)

p value: <0.0001
(Pearson's Chi-squared test)

Firearms

636 (78.0%)

179 (22.0%)

815 (1.5%)

Graffiti

490 (49.9%)

492 (50.1%)

982 (1.8%)

Volatile substance

28 (68.3%)

13 (31.7%)

41 (0.1%)

Weapons

3099 (67.2%)

1512 (32.8%)

4611 (8.6%)

Total

44189 (82.6%)

9327 (17.4%)

53516 (100.0%)

label

variable

Found

Total

test

No

Yes

Year

2018

10566 (82.3%)

2275 (17.7%)

12841 (24.4%)

p value: 0.0033
(Pearson's Chi-squared test)

2019

10915 (82.2%)

2366 (17.8%)

13281 (25.2%)

2022

12534 (83.6%)

2453 (16.4%)

14987 (28.5%)

2023

9587 (83.0%)

1958 (17.0%)

11545 (21.9%)

Total

43602 (82.8%)

9052 (17.2%)

52654 (100.0%)

Unit.type

CIU

1695 (85.6%)

284 (14.4%)

1979 (3.8%)

p value: <0.0001
(Pearson's Chi-squared test)

DRU

1536 (88.4%)

201 (11.6%)

1737 (3.3%)

Highway Patrol

374 (66.1%)

192 (33.9%)

566 (1.1%)

Other

932 (85.3%)

161 (14.7%)

1093 (2.1%)

PSO

639 (66.7%)

319 (33.3%)

958 (1.8%)

Public Order Response

4018 (86.0%)

655 (14.0%)

4673 (8.9%)

Transit

2821 (75.1%)

934 (24.9%)

3755 (7.1%)

Uniform

31587 (83.4%)

6306 (16.6%)

37893 (72.0%)

Total

43602 (82.8%)

9052 (17.2%)

52654 (100.0%)

Gender

F

5999 (86.1%)

966 (13.9%)

6965 (18.8%)

p value: <0.0001
(Fisher's Exact Test for Count Data)

M

24873 (82.6%)

5233 (17.4%)

30106 (81.2%)

U

17 (100.0%)

0 (0%)

17 (0.05%)

NA

12713

2853

15566

Total

43602 (82.8%)

9052 (17.2%)

52654 (100.0%)

Area.type

Metro

23851 (83.9%)

4572 (16.1%)

28423 (73.2%)

p value: <0.0001
(Pearson's Chi-squared test)

Regional

8539 (82.1%)

1862 (17.9%)

10401 (26.8%)

NA

11212

2618

13830

Total

43602 (82.8%)

9052 (17.2%)

52654 (100.0%)

Comparison of hit rates for different ethnic appearances by LGA

Searches that were undertaken by an officer based at a local Police station were able to be mapped to their Local Government Area. The hit rates for persons of different ethnic appearances for Local Government Areas appears below.

Where there were less than five searches for a given group in a Local Government Areas hit rates are not shown, as statistical results may be misleading.